small image Ebook: Apache NiFi for Dummies

Impact

Increased customer ratings through improved service

Identified 17x more fraud

Delivered platform at 1/10 the cost of traditional data warehouse and BI environment

Challenges

Gain an end-to-end view of the customer experience

Support an increasing volume of telemetry data

More rapidly uncover new fraud tactics

Solutions

Cloudera Enterprise

Apache Kafka, Apache Spark

Industry

Telecommunications

Applications supported

Data discovery and analytics for service monitoring, fraud monitoring, and customer care; machine learning for real-time processing, troubleshooting, and analysis; stream processing for real-time alerting

Big Data Scale

Nearing 1PB

Cisco WebEx, a leader in conferencing services, improved its customer ratings and uncovered up to 17 times more fraud when it moved from disparate data silos to a data discovery and analytics environment on Cloudera Enterprise and Cisco UCS Servers.

Overview

Cisco WebEx supports more than 26 billion conference minutes each month. Its audio, video, and web conferencing services help users connect and collaborate with colleagues around the world.

Maintaining a leadership position in the web conferencing market requires keeping a keen eye on service quality and customer needs. The time it takes to join a meeting, the quality of the audio and video feeds, and the ability for users to easily collaborate all play a role in customer satisfaction and ratings. However, one of the challenges Cisco WebEx faced was that different data silos were used to track service issues, client usage, and potential fraud, and staff didn’t have an easy way to combine or correlate data across these silos. This limited the company’s ability to see end-to-end the customer experience. Additionally, the company’s current tracking couldn’t support an increasing volume of telemetry data—expected to rise to several petabytes in the next few years.

Solution

Cloudera Enterprise, Platfora Big Data Discovery, and Cisco Unified Computing System (UCS) servers drive an Apache Hadoop based enterprise data hub, providing a platform that can scale out to accommodate petabytes of data at one-tenth the cost of a traditional data warehousing and BI infrastructure.

Cloudera and Platfora enabled us to significantly reduce the number of tools we need to manage, and decreased the time to launch new use cases from one month to less than a day.

–Joe Hsy, director, Development and Engineering, Cisco WebEx

By moving from disparate data silos to a data discovery and analytics environment on this enterprise data hub, WebEx staff can now combine and correlate operational data with business information to improve decision-making across all aspects of its business—product development, engineering, marketing, sales, and customer service. Machine-learning algorithms in Apache Spark enable the company to automatically create new rules as new patterns are detected to continuously improve fraud detection and service monitoring. The unified management capabilities of Cisco UCS servers reduce provisioning times by close to 80 percent.

Impact: Improved ratings, reduced fraud

One of the team’s first use cases for its new platform was understanding “meeting join time”—how long it takes a customer to join a meeting—as this has a direct impact on the user’s overall meeting experience.

With its new data discovery and analytics environment, staff can drill down into all the necessary data—web data, server data, network data, user behavioral data, back-end system data—to understand not only that a problem is occurring, but why. Such insight has enabled the organization to better isolate the root cause of problems and take action to improve the customer experience.

With Cloudera and Platfora, we've seen visible improvement in terms of the survey data we have. Our customers are happier, which is ultimately our goal.

–Joe Hsy, director, Development and Engineering, Cisco WebEx

Additionally, using machine-learning algorithms, the system can identify new fraud patterns as they evolve and automatically create the new rule sets to keep up with fraudsters’ evolving tactics. Previously, the organization had to manually code “rules” based on detected fraud patterns, and while the manual rules helped identify many fraudulent cases, it was difficult for staff to keep up with fraudsters’ evolving tactics.

When we compared both approaches, the machine-learned rules did much better, helping us detect up to 17 times more fraud based on historic data.

Joe Hsy, director, Development and Engineering, Cisco WebEx

Your form submission has failed.

This may have been caused by one of the following:

  • Your request timed out
  • A plugin/browser extension blocked the submission. If you have an ad blocking plugin please disable it and close this message to reload the page.